[erlang-questions] [ANN] misultin v0.7
Roberto Ostinelli
roberto@REDACTED
Sun Apr 10 21:10:48 CEST 2011
On Apr 10, 2011, at 6:47 PM, Max Lapshin wrote:
It seems to be an eternal process: new webserver which takes not 50
microseconds, but 10 of them per request and is 5 times faster becomes
the same mochiweb when it reaches it by functionality.
[...]
And now I can tell, that there are some features, that are much more
important that exact number of _micro_seconds per request: it is
memory leakage.
i can only agree to that. stability and functionalities [the ones you need!]
are the most important part of all.
it is quite normal that a library slows down as features get added, such as
limiting the data that malicious client can send, etc.
On Apr 10, 2011, at 7:36 PM, Bob Ippolito wrote:
I say slightly slower because the last time I benchmarked it against
misultin, it only differed by a few percent. For most purposes, this
is an irrelevant difference.
i slightly disagree on this, since these are my latest tests with jmeter
comparing mochiweb 1.5.2 tag with misultin master tree [convention is name
of library / concurrent threads], over a simple echoing of a variable passed
in GET:
library & threads # samples Avg (ms) Min (ms) Max (ms)
StdDev Error Throughput /sec KB/sec Avg. Bytes
Mochi 1 100000 0 0 330 1.11 0 1129.80 54.06 49
Mochi 10 100000 4 0 36 1.30 0 2221.83 106.32 49
Mochi 50 100000 20 0 148 4.97 0 2393.49 114.53 49
Mochi 100 100000 39 1 335 10.04 0 2436.41 116.59
49
Mochi 200 100000 83 1 381 23.27 0 2321.53 111.09
49
Misultin 1 100000 0 0 38 0.53 0 1545.26 73.94 49
Misultin 10 100000 2 0 318 1.33 0 3548.24 169.79
49
Misultin 50 100000 12 0 46 3.68 0 3899.70 186.60
49
Misultin 100 100000 24 0 110 7.36 0 3928.66 187.99
49
Misultin 200 100000 55 1 223 18.58 0 3477.29
166.39 49
however, as i've always said, mochiweb is a wonderful library which as a
proven and successful history in running in production. misultin is used in
production too, but i wouldn't be surprised that it is less stable.
i've spent much less time in optimizing speed than in putting efforts to
functionalities and such. there's a perfectly good reason for that. :)
r.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20110410/52a962ea/attachment.htm>
More information about the erlang-questions
mailing list